Less is more: Efficient behavioral context recognition using Dissimilarity-Based Query Strategy

• DOI: 10.1371/journal.pone.0286919
• Publication Date: 2023-06-07T17:29:20Z
• Authors: Atia Akram; Asma Ahmad Farhan; Amna Basharat
• Keywords: Activity Recognition; Intuition

With the advancement of ubiquitous computing, smartphone sensors are generating a vast amount unlabeled data streams ubiquitously. This sensor can potentially help to recognize various behavioral contexts in natural environment. Accurate context recognition has wide variety applications many domains like disease prevention and independent living. However, despite availability enormous amounts data, label acquisition, due its dependence on users, is still challenging task. In this work, we propose novel approach i.e., Dissimilarity-Based Query Strategy ( DBQS ). Our leverages Active Learning based selective sampling find informative diverse samples train model. overcomes stagnation problem by considering only new distinct from pool that were not previously explored. Further, our model exploits temporal information order further maintain diversity dataset. The key intuition behind proposed variations during learning phase will settings it outperform when assigned task setting. Experimentation publicly available environment dataset demonstrates improved overall average Balanced Accuracy(BA) 6% with an 13% less training requirement.